A Forward-Looking Exploration: Developing a Computer-aided Tool for Ocular Symptom Assessment in Myasthenia Gravis Based on Facial Images

Myasthenia Gravis (MG) is an autoimmune neuromuscular disorder that heavily affects various daily actions of the suffers. The manifestations of MG disease are fluctuating muscle weakness and fatigue across different muscle units of human body, while ocular muscles involvement is regarded as primary symptom in a majority of the patients. Currently, the conventional clinical approach for MG assessment involves identifying and measuring key ocular structures manually, which is resource-intensive and prone to subjective errors. In our paper, a novel application of computer vision technique is introduced to assist in the quantitative evaluation of extraocular muscles in MG patients. By employing advanced segmentation algorithms, both prior feature-driven and deep feature-driven methods, we aims to automatically determine the ocular structures such as the iris, sclera, and pupil, etc. This pilot study serves as a proof of concept for using classical segmentation models to facilitate automatic and non-invasive MG assessments, enhancing clinical management and patient autonomy in monitoring their condition. Our approach not only aids physicians in obtaining more accurate measurements for better disease management but also reduces the necessity for frequent clinical visits, potentially improving adherence to treatment and overall quality of life for MG patients.